It is known that individuals' personal networks significantly affect their physical and mental well-being. Yet much less is known - and much less reliably - how those networks change over time, respond to life events, and are reconstructed by individuals, which means that the understanding of how networks affect health remains significantly limited. This project is designed to address three major issues. First, the proposed project will be a unique panel study generating a rich picture of network structure (rather than summary assessments by respondents) via multiple name-eliciting questions in a large sample, adding now the dimension of time. Second, these data will reveal the dynamics of personal networks: how they change, what makes them change, and how individuals adapt to those changes. The study results will indicate which individuals are better- or worse-equipped to respond to events and to sustain their networks - and with what health consequences. No existing American published studies or on-going data-collecting surveys address such issues by combining rich network data, large samples, and a panel structure. Third, the proposed project will conduct experiments on the network-measuring methodology to enable effective studies of other samples as well, in particular, by comparing data gathered in face-to-face interviews with data gathered over the internet. This project will describe, track, and explain network change and reconstruction. It goes far beyond existing studies by combining (1) in-depth, multidimensional, and reliable measures of personal networks with (2) a panel design, in (3) a sufficiently large, yet focused, sample. Specifically, the proposed study consists of three waves of interviews, 18 months apart, of 21-30 and 50-to-70 year-olds, using the bulk of the interview time in each wave to elicit detailed descriptions of their personal social ties. These questions wil yield many multidimensional measures of personal networks. The remainder of the interview time will be devoted to measuring physical and mental health, life events, and personality. The administration will entail systematically mixing face-to-face and web interviewing over the three waves in a design that will allow evaluations of mode effects on network (and other) measures, and development of an effective web-based instrument for future research on personal networks.